| Author | Thread |
|
|
04/20/2006 03:12:25 PM · #1 |
| What if eventualy the sensors become so large they need a larger appature? is it possible? is 1 as large as they can go? will the numbers ever get into the 0.5 or into the negative? |
|
|
|
04/20/2006 03:17:26 PM · #2 |
No, 1 is not as large as they can go. Thef-stop is a ratio. It's the focal length divided by the aperture size, so if the aperture is bigger than the focal length, the result can be less than one. It cannot, however, be negative.
There have been lenses as fast as f/0.95, and Canon used to produce a 50/1.0L, but the DoF is incredibly thin for these uber-fast lenses and the optical performance leaves something to be desired. Don't look for sub-f/1.0 lenses to appear with any regularity in the future.
|
|
|
|
04/20/2006 03:33:24 PM · #3 |
Originally posted by BowerR64:
What if eventualy the sensors become so large they need a larger appature? is it possible? is 1 as large as they can go? will the numbers ever get into the 0.5 or into the negative? |
Interestingly enough, that's one of the major tradeoffs in the film world; as you move up to larger and larger formats, your "normal" lens getslonger and longer; a 50mm is a "normal" field of view on 35mm cameras, but for 4x5 film you need a 150 mm to cover the same field of view, and something like a 210mm on 8x10 film. And since for optical reasons alone there's a practical limit in how "large" an f/stop can be (as Kirbic pointed out), expecially on the longer focal lengths, this means that moving up in film/sensor size means giving up raw speed in the lens to some degree, for a given coverage, or at the very least dramatically increasing the COST of attaining that speed.
For example; my Canon 70-200mm f/4L lens is effectively a 112-320mm f/4 lens on the 20D's APS-C sensor. If I were using the 5D camera, I'd have to buy the Canon 100-400mm f/4.5-5.6L lens to get roughly the equivalent coverage. This would be a slower lens, not constant aperture, and the COST difference (not to mention size and weight) is dramatic: roughly $600 vs roughly $1,300...
Add to this the fact that as you move to longer and longer lenses for a given angular coverage on larger sensors, you also have to stop them down to a smaller f/stop to get the same DOF, and you'll see that there are distinct disadvantages to working with full-frame, or even larger, sensors, and that they definitely are not a holy grail to which everyone should be aspiring, marketing pushes and popular misinformation notwithstanding.
R.
Message edited by author 2006-04-20 16:38:25.
|
|
|
|
04/20/2006 04:30:16 PM · #4 |
Originally posted by Bear_Music:
Originally posted by BowerR64:
What if eventualy the sensors become so large they need a larger appature? is it possible? is 1 as large as they can go? will the numbers ever get into the 0.5 or into the negative? |
Interestingly enough, that's one of the major tradeoffs in the film world; as you move up to larger and larger formats, your "normal" lens getslonger and longer; a 50mm is a "normal" field of view on 35mm cameras, but for 4x5 film you need a 135 mm to cover the same field of view, and something like a 210mm on 8x10 film. And since for optical reasons alone there's a practical limit in how "large" an f/stop can be (as Kirbic pointed out), expecially on the longer focal lengths, this means that moving up in film/sensor size means giving up raw speed in the lens to some degree, for a given coverage, or at the very least dramatically increasing the COST of attaining that speed.
For example; my Canon 70-200mm f/4L lens is effectively a 112-320mm f/4 lens on the 20D's APS-C sensor. If I were using the 5D camera, I'd have to buy the Canon 100-400mm f/4.5-5.6L lens to get roughly the equivalent coverage. This would be a slower lens, not constant aperture, and the COST difference (not to mention size and weight) is dramatic: roughly $600 vs roughly $1,300...
Add to this the fact that as you move to longer and longer lenses for a given angular coverage on larger sensors, you also have to stop them down to a smaller f/stop to get the same DOF, and you'll see that there are distinct disadvantages to working with full-frame, or even larger, sensors, and that they definitely are not a holy grail to which everyone should be aspiring, marketing pushes and popular misinformation notwithstanding.
R. |
As an aside, a "normal" lens is typically thought of as a lens with a focal length equal to the diagonal measurement of the frame. So, for a 35mm camera, "normal" works out to 43mm, for 4x5 "normal" is 162mm and for 8x10 it's 325mm.
The f-stop could theoretically approach zero, but to actually attain an fstop of zero would require an aperture of infinite size. The f-stop can never be less than zero, because you cannot have a physical dimension with a magnitude that is less than zero.
From a practical standpoint, I doubt that we'll see many lenses with apertures less than 1.2 or so anytime soon. (Yes, I know, there are exceptions that already exist, but they are rarities) Such lenses are really not terribly practical for photography as the DOF gets ridiculously narrow.
Message edited by author 2006-04-20 19:11:55. |
|
|
|
04/20/2006 04:42:30 PM · #5 |
Originally posted by Spazmo99:
As an aside, a "normal" lens is typically thought of as a lens with a focal length equal to the diagonal measurement of the frame. So, for a 35mm camera, "normal" works out to 43mm, for 4x5 "normal" is 162mm and for 8x10 it's 325mm. |
That's a little more precise, thanks. I actually meant to enter 150mm for 4x5 "normal" but somehow ended up with 135 instead :-( I've fixed it in my original post. I've no doubt you're right about the 325mm being "statutory normal" for 8x10, I was just going by my memory that we considered the 210 to be our standard lens in that format. Probably just because with the really large format the tendency is to want to go wider and have a little cropping room in the image circle for camera movements; understandably, the movement-before-vignetting is pretty limited on an 8x10 compared with a 4x5, so in architectural photography we often had to compose wider and crop out foreground, basically.
R.
|
|
|
|
04/20/2006 04:49:17 PM · #6 |
Damn you people are smart. It'll take me a week of studying now to figure out what was said here. Thanks Robert for the info! Now if I can just get it to sink in.
|
|
|
|
04/20/2006 04:54:06 PM · #7 |
Originally posted by kirbic:
No, 1 is not as large as they can go. Thef-stop is a ratio. It's the focal length divided by the aperture size, so if the aperture is bigger than the focal length, the result can be less than one. It cannot, however, be negative.
There have been lenses as fast as f/0.95, and Canon used to produce a 50/1.0L, but the DoF is incredibly thin for these uber-fast lenses and the optical performance leaves something to be desired. Don't look for sub-f/1.0 lenses to appear with any regularity in the future. |
Didn't Stanley Kubrik use a f/0.7 lens to shoot the interior shots in 'Barry Lyndon'?
|
|
|
|
04/20/2006 05:03:40 PM · #8 |
| Here is an interesting informationIt contains information about Zeiss 50mm f0.7 lenses, Stanley Kubric and other trivia. |
|
|
|
04/20/2006 05:12:24 PM · #9 |
Can the appature be larger then the sensor? or is it already?
Why cant you have a square lense? or why isnt the sensor round?
Sorry if these are dumb, just thing ive always wondered. Why lenses are round but the pictures are developed square. |
|
|
|
04/20/2006 05:15:26 PM · #10 |
You can't have square lenses. If you did have square lenses and you dropped your camera bag, no one could laugh at you as you chased all of your rolling lenses into the street :-)
Originally posted by BowerR64:
Can the appature be larger then the sensor? or is it already?
Why cant you have a square lense? or why isnt the sensor round?
Sorry if these are dumb, just thing ive always wondered. Why lenses are round but the pictures are developed square. |
|
|
|
|
04/20/2006 06:50:46 PM · #11 |
Originally posted by BowerR64:
Can the appature be larger then the sensor? or is it already?
Why cant you have a square lense? or why isnt the sensor round?
Sorry if these are dumb, just thing ive always wondered. Why lenses are round but the pictures are developed square. |
Lenses actually "throw" circular images. Sensors (or film) crop out a portion of that image for you. Whether or not a given sensor (or film) id large enough to containt he entire image projected by the lens is a matter of "format" (large or small).
With lenses that are designed specifically to a particular format, like (say) lenses for 35mm cameras, the image circle that is thrown is of such a size that when you place the film within the circle and centered upon it, the corners of the film are very nearly touching the limits of the circle.
But, as stated earlier, a 150mm lens is "normal" for a 4x5 inch film, whereas a 43mm (or 50mm) is "normal" for a 35mm film. The difference is that a 150mm lens designed for large format cameras throws a MUCH larger image circle that covers a wider angle than the artifically-cropped circle of the 35mm-specific 150 mm lens.
If you mount a 35mm film to the film plane of a view camera with a 150mm lens on it, the film will "read" only a very small portion of the image circle, and the resultant "cropped" image will be identical to the one you'd have if you mounted a 150mm lens on a 35mm camera body.
So, in practical terms, focal length is meaningless except when it is correlated with sensor/film size. This is why we have the oft-mentioned "crop factor" in the dSLR world. The 35mm film/focal length relationship has become accepted as a de facto standard, and we have taken to "converting" focallengths into "equivalents" so we can compare apples with oranges, so to speak.
It's very confusing, I know. If you can ever go someplace where they are using a "view camera" (large format camera) then take the opportunity, and ask them to show you what I'm talking about. When you see how "camera movements" are used to place the film precisely where you need it within the image circle, suddenly image circles will start to make sense.
Incidentally, a 50mm f/2 lens has a 25mm diameter aperture when wide open, and there are certainly sensors smaller than 25mm, so yes it's definitely possible to have an aperture larger than a sensor. But that would be a VERY big lens on a VERY small camera, and I'm not sure such a combo exists. In the real world, as sensors get smaller focal lengths get shorter (my 10mm lens on the APS-C Canon sensor is the same in terms of coverage, as Kirbic's 16mm lens on his FF sensor), and since f/stop is a ratio between the diameter of the aperture and the focal length of the lens, as the lenses get shorter the physical diameter of the aperture at a given f/stop is correspondingly smaller, and voila: the whole lens is smaller and lighter. And cheaper :-)
Robt.
|
|
|
|
04/21/2006 02:04:47 AM · #12 |
Does it matter where the appature is in the lens?
You look at the shorter lenses andit appears the appature is in the center but when you look at a longer lens it seems to be in the center also. If it was closer to the sensor wouldnt that make it a little larger?
Can you polish bad glass and make it better? older glass that looks foggy. |
|
|
|
04/21/2006 02:35:54 AM · #13 |
if only there are round shaped sensors.
Then all the light will not go wasted, lol |
|
|
|
04/21/2006 02:40:18 AM · #14 |
Originally posted by crayon:
if only there are round shaped sensors.
Then all the light will not go wasted, lol |
But think of all the wasted PAPER until they come up with round printers and round frames :-)
R.
|
|
|
|
04/21/2006 03:04:52 AM · #15 |
Originally posted by Bear_Music:
Originally posted by crayon:
if only there are round shaped sensors.
Then all the light will not go wasted, lol |
But think of all the wasted PAPER until they come up with round printers and round frames :-) |
hehe of course, you are right-on there, Robt :p
OK, just another crazy thought. What if the world started off with circles instead of squares/rectangles as what we are used to, today? What if all along, things had always been circles? Our TV has a circle tube, our computers have a circle screen, our mobile phone screens, photo frames and everything else - circles?
I think the ironic thing might be that, the digital camera lens and optics would then be rectangular, ROTFLMAO! 
Message edited by author 2006-04-21 03:05:44. |
|
|
|
04/21/2006 03:06:55 AM · #16 |
Originally posted by crayon:
Originally posted by Bear_Music:
Originally posted by crayon:
if only there are round shaped sensors.
Then all the light will not go wasted, lol |
But think of all the wasted PAPER until they come up with round printers and round frames :-) |
hehe of course, you are right-on there, Robt :p
OK, just another crazy thought. What if the world started off with circles instead of squares/rectangles as what we are used to, today? What if all along, things had always been circles? Our TV has a circle tube, our computers have a circle screen, our mobile phone screens, photo frames and everything else - circles?
I think the ironic thing might be that, the digital camera lens and optics would then be rectangular, ROTFLMAO! |
Sleep it off; things will be back to normal in the morning...
R.
|
|
|
|
04/21/2006 01:08:27 PM · #17 |
Originally posted by crayon:
Originally posted by Bear_Music:
Originally posted by crayon:
if only there are round shaped sensors.
Then all the light will not go wasted, lol |
But think of all the wasted PAPER until they come up with round printers and round frames :-) |
hehe of course, you are right-on there, Robt :p
OK, just another crazy thought. What if the world started off with circles instead of squares/rectangles as what we are used to, today? What if all along, things had always been circles? Our TV has a circle tube, our computers have a circle screen, our mobile phone screens, photo frames and everything else - circles?
I think the ironic thing might be that, the digital camera lens and optics would then be rectangular, ROTFLMAO! |
The first (or at least one of the first) TVs did use a circular display. |
|
|
|
04/21/2006 01:11:07 PM · #18 |
Originally posted by BowerR64:
Can the appature be larger then the sensor? or is it already?
Why cant you have a square lense? or why isnt the sensor round?
Sorry if these are dumb, just thing ive always wondered. Why lenses are round but the pictures are developed square. |
What I wonder is why aren't DSLR sensors a 36mm x 36mm square? That way, you can use your existing lenses, and you could crop horizontal, vertical or a larger square image. (Of course, you'd need some new hoods.) |
|
|
|
04/21/2006 01:23:43 PM · #19 |
Originally posted by Bear_Music:
For example; my Canon 70-200mm f/4L lens is effectively a 112-320mm f/4 lens on the 20D's APS-C sensor. If I were using the 5D camera, I'd have to buy the Canon 100-400mm f/4.5-5.6L lens to get roughly the equivalent coverage. This would be a slower lens, not constant aperture, and the COST difference (not to mention size and weight) is dramatic: roughly $600 vs roughly $1,300...
|
Another way of looking at this is that for a given lens length, the 20D crops the photo for you, but you have the ability to crop it yourself on the 5D (you lose some resolution though). In other words, you can get crops of 1.6 and higher on the 20D, and crops of 1.0 and higher on the 5D. (a beter comparison would be the 1DMII and the 5D, which have the same pixel pitch).
Originally posted by Bear_Music:
Add to this the fact that as you move to longer and longer lenses for a given angular coverage on larger sensors, you also have to stop them down to a smaller f/stop to get the same DOF, and you'll see that there are distinct disadvantages to working with full-frame, or even larger, sensors, and that they definitely are not a holy grail to which everyone should be aspiring, marketing pushes and popular misinformation notwithstanding. |
Bear_Music didn't mention it, but you can't go to an infinitely high f/stop becasue of difraction issues. Diffraction is a complex phenomena that deserves its own thread (and there are plenty of them).
If you remember back to college or even high school physics class, you probably did a wave-tank experiment where you saw water waves propagate (travel) as strait line until they hit a barier with a small hole (the aperature) in it. Past that hole, the waves expanded in a semi-circle. Light behaves pretty much the same way, except that there are several frequecies of light waves which interfere with each other and light waves propaget in three dimensions, which makes the whole issue much more complex.
Message edited by author 2006-04-21 13:24:17. |
|
|
|
04/21/2006 01:28:40 PM · #20 |
Originally posted by hankk:
Bear_Music didn't mention it, but you can't go to an infinitely high f/stop becasue of difraction issues. Diffraction is a complex phenomena that deserves its own thread (and there are plenty of them).
If you remember back to college or even high school physics class, you probably did a wave-tank experiment where you saw water waves propagate (travel) as strait line until they hit a barier with a small hole (the aperature) in it. Past that hole, the waves expanded in a semi-circle. Light behaves pretty much the same way, except that there are several frequecies of light waves which interfere with each other and light waves propaget in three dimensions, which makes the whole issue much more complex. |
This is true, but we're specifically discussing LARGE apertures here, not small ones :-)
R.
|
|
|
|
04/21/2006 01:33:54 PM · #21 |
Originally posted by hankk:
What I wonder is why aren't DSLR sensors a 36mm x 36mm square? That way, you can use your existing lenses, and you could crop horizontal, vertical or a larger square image. (Of course, you'd need some new hoods.) |
Speaking in terms of full-frame sensors like the 5D has, if you wanted a square image you'd have to decrease the long dimension. Draw a circle and place in it a rectangle that's both centered and tangent on all 4 corners; the longer you make the long dimension, the shorter you must make the short dimension for the whole to remainw ithin the image circle. At the extreme, the longest dimension you could fit in would be equal to the diameter of the circle, but the other dimension would have to be "0" if the whole were not to vignette.
Of course, you could theoretically float a square APS-C size sensor within the image circle of a full-frame lens...
R.
|
|
|
|
04/21/2006 02:22:10 PM · #22 |
Couple things from above...
- There are LOTS of lenses where the aperture is larger than the sensor, even for full-frame 35mm sensors which have diagonals of about 43mm. Examples:
Canon 100 f/2, 50mm aperture
Canon 200 f/2.8, 71mm aperture
Canon 85 f/1.2L, 85/1.2=71mm aperture
Canon 300 f/2.8, 107mm aperture
Canon 200 f/1.8, 111mm aperture
Canon 400 f/2.8, 142mm aperture
Canon 600 f/4, 150mm aperture
From the above you'll notice that as the focal length gets longer, apertures need to be very large for faster lenses. When we talk about "aperture size" many asume that it's the physical dimension of the diaphragm opening. That's not the case. It's actually the diameter of the incoming cylinder of light rays, and is approximated by the *usable* diameter of the front element, at least for longer focal lengths.
With regard to square vs. rectangular sensors, you'd think that a square sensor would capture much more of the image circle, but it's barely an advantage. In fact, there's only an 8% advantage in area for a square sensor vs. a 1:1.5 aspect ratio sensor!
|
|
|
|
04/21/2006 03:05:31 PM · #23 |
Uhm.... my head's going to explode.
 |
|
|
|
04/21/2006 03:19:50 PM · #24 |
Originally posted by Bear_Music:
Originally posted by hankk:
What I wonder is why aren't DSLR sensors a 36mm x 36mm square? That way, you can use your existing lenses, and you could crop horizontal, vertical or a larger square image. (Of course, you'd need some new hoods.) |
Speaking in terms of full-frame sensors like the 5D has, if you wanted a square image you'd have to decrease the long dimension. Draw a circle and place in it a rectangle that's both centered and tangent on all 4 corners; the longer you make the long dimension, the shorter you must make the short dimension for the whole to remainw ithin the image circle. At the extreme, the longest dimension you could fit in would be equal to the diameter of the circle, but the other dimension would have to be "0" if the whole were not to vignette.
Of course, you could theoretically float a square APS-C size sensor within the image circle of a full-frame lens...
R. |
You could always just make a cross shaped sensor then - capture the landscape and portrait max area in a single frame. Though a cross doesn't tesselate so easily for wafer fab as a square or rectangualar sensor array.
Though, then while you were at it, you may as well fill in the corners of the cross and allow for better rotation tolerance for fixing horizons and not 'loosing' pixels in the corners...
The reason most of this isn't done though is cost. More sensor area is directly related to the cost of manufacture. More pixels increase the likelyhood of a failed sensor (which is mostly a per area failure %)
Creating 'extra' to cover more of the lens circle increases the cost with little advantage. Though my P&S has a 16:9 sensor, with lens options to crop to different ratios in camera, which is quite neat. Nothing I couldn't do in photoshop later, but helps with the compositional thought to do it there and then.
Message edited by author 2006-04-21 15:21:18. |
|
|
|
04/21/2006 03:57:09 PM · #25 |
Gordon's thought exercise imagining a cross-shaped sensor and then filling in the ccorners to make a square is interesting. It results in a 36mm square sensor (of course) that covers 78.6% of a 43.2mm image circle, vs. just 63.6% for a 30.55mm square (the 30.55mm square is the best efficiency you can get with a diagonal size of exactly 43.2mm). That's a whopping gain in image circle coverage, but with a stiff penalty... it's area is 1.5x greater than a 24x36mm sensor, and semiconductor production costs per unit are are relatively constant. 1.5x doesn't seem to be TOO bad a cost penalty, but we haven't considered yield, which falls rapidly as size increases. Canon's yield on 24x36mm sensors was 10% as recently as a couple years ago; it's above 25% now, and will likely improve dramatically in the near future, but for now this size sensor would be exorbitantly expensive.
All that, and you'd have to crop anyway, LOL. The corners would be outside the image circle and thus would be dark.
|
|
